Print and scan versus software: which is actually safer?
Ask how to black something out properly and you will meet two camps who are both certain. One says print it, cover it with a marker, and scan it back in, because paper cannot leak what is not on it. The other says that is unprofessional, lossy and unnecessary when software can do the job.
Both have a real argument. The disagreement is rarely laid out honestly, so here it is.
The case for print and scan
It is not a folk remedy. The United States Court of Federal Claims recommends exactly this in its own guidance: print the document, obscure the material with a marker, and scan the result back in as the filed version.
The reasoning is that it removes an entire class of failure by leaving the digital file behind. A printed page carries no text layer, no metadata, no earlier revision, no bookmarks, no attachments and no form fields. Every one of the eight things the copy-paste test misses is a property of a file, and you have stopped having that file. What the scanner produces is a photograph of a piece of paper with ink on it.
It also needs no software and no trust. Anyone can do it with equipment they already have, and nobody has to believe a vendor's claim about what a menu item does.
The case against it
Four objections, and three of them are practical rather than snobbery.
- Markers are more translucent than they look. A single pass of a black marker over laser print often leaves the letters visible when the page is held to the light or scanned at a high setting, because the ink sits on top of toner rather than soaking through it. Scanners are better at seeing through it than your eyes are.
- Your scanner may put the text straight back. Many office scanners and scanning apps run text recognition by default and write an invisible, searchable text layer underneath the picture. It reads the page as it now appears, so it will not recover what the marker covered — but if the marker failed to cover it completely, the recognition may read it perfectly well and write it into the file as text. The failure and the fix arrive together.
- Quality drops and it shows. The result is visibly a scan of a print: heavier, skewed, sometimes grey. For a document going to a letting agent this is fine. For a document going into a professional file it looks like what it is.
- It does not scale. Two pages is a minor chore. Sixty pages is an afternoon, and the chance of missing one grows with every page.
What the comparison actually turns on
The two options are not really print against software. They are removing the text against covering the text, and both methods can do either.
| Method | Are the words still in the file? | What can still go wrong |
|---|---|---|
| Highlighter or shape in a word processor | Yes, always | Everything. This is the failure in most of the published cases. |
| Print, marker, scan | No | Marker too thin; the scanner writing a fresh text layer; missed pages; quality |
| PDF software with a genuine redaction function | No, when it works | Metadata, earlier versions, bookmarks, thumbnails, attachments and form fields are separate from the page and are not always cleared |
| Rebuilding each page as an image | No | The output cannot be searched or read aloud; a box that lands wrong is still a box that lands wrong |
Print and scan is a manual version of the last row. It rebuilds the page as a picture using a printer and a scanner instead of code, and it inherits the same strength and the same weakness.
Where we come down
For one or two pages, with no scanner that runs text recognition, print and scan is sound advice and the court that recommends it is not being old-fashioned. Use a wide marker, go over every line twice, and hold the page up to a lamp before you scan it.
For anything longer, or where the document has to stay legible, doing it in software is better — on one condition, which is the condition this whole argument usually skips. You have to check the output. Print and scan gives you confidence you can reason about, because you can see the paper. Software gives you no such feedback, and a covered word and a removed word look exactly the same on screen.
So the honest comparison is not between two methods. It is between a method you can verify and one you cannot, and software only wins once you add the check that makes it verifiable.
Either way, check the file
Check a document — free, no account, nothing uploaded. It works on a scan as well as on a born-digital file, and on a scan it will tell you whether a text layer was written underneath the picture.
Nothing here is legal advice, and no tool makes anyone compliant with anything.